A prolonged MR T2 relaxation time was proposed to mark the presence and severity of Alzheimer's disease (AD). We studied the value of T2 relaxometry in diagnosing early AD. T2 was measured from 54 patients with AD, 25 subjects with age-associated memory impairment (AAMI), 18 elderly and 16 young controls. The AD patients had longer T2 in the right hippocampal head (104 +/- 11 ms) and tail (98 +/- 10 ms) than age-matched controls (95 +/- 5 and 92 +/- 9 ms, respectively). This prolongation was not related to age. In the AD group, the T2 of the left hippocampal head also correlated with the clinical severity. The T2 of the amygdala did not differ across the groups. Increased T2 in the temporal and parietal white matter and the thalamus related to increasing age rather than to the diagnostic category. The AAMI subjects had T2 comparable with those of age-matched controls. Despite the prolongation of T2 in the AD group the possible diagnostic value was compromized by a substantial overlap between the study groups. We, thus, conclude that the T2 relaxometry is not a reliable method for diagnosing early AD.